A Model Combining Stacked Auto Encoder and Back Propagation Algorithm for Short-Term Wind Power Forecasting

Abstract

Recently, many countries have spent great efforts on wind power generation. Although there have been many methods in the field of wind power forecasting, the persistence statistics model based on historical data is still being challenged due to the randomness and uncontrollability in wind power. Hence, a more accurate and effective wind power forecasting method is still required. In this paper, a new forecasting method is proposed by combining stacked auto-encoders (SAE) and the back propagation (BP) algorithm. First, an SAE with three hidden layers is designed to extract the characteristics from the reference data sequence, and the subsequent loss function is used in the pre-training process to obtain the optimal initial connection weights of the deep network. Second, after adding one output layer to the stacked auto encoders, the BP algorithm is used to fine tune the weights of the whole network. To achieve the best network architecture, the particle swarm optimization is adopted to decide the number of neurons of the hidden layer and the learning rate of each auto encoder. Experimental results show that, for short-term wind power forecasting, the proposed method achieves more stable and effective performance than the existing BP neural network and support vector machines. The improvement in accuracy is 12% on average under different time steps.

Keywords

• Machine learning
• particle swarm optimization
• stacked auto-encoders
• wind energy
• wind power forecasting